|Home | About | Journals | Submit | Contact Us | Français|
In this study, the feasibility of fluorescence lifetime imaging (FLIM) for measurement of RNA:DNA ratios in microorganisms was assessed. The fluorescence lifetime of a nucleic acid-specific probe (SYTO 13) was used to directly measure the RNA:DNA ratio inside living bacterial cells. In vitro, SYTO 13 showed shorter fluorescence lifetimes in DNA solutions than in RNA solutions. Growth experiments with bacterial monocultures were performed in liquid media. The results demonstrated the suitability of SYTO 13 for measuring the growth-phase-dependent RNA:DNA ratio in Escherichia coli cells. The fluorescence lifetime of SYTO 13 reflected the known changes of the RNA:DNA ratio in microbial cells during different growth phases. As a result, the growth rate of E. coli cells strongly correlated with the fluorescence lifetime. Finally, the fluorescence lifetimes of SYTO 13 in slow- and fast-growing biofilms were compared. For this purpose, biofilms developed from activated sludge were grown as autotrophic and heterotrophic communities. The FLIM data clearly showed a longer fluorescence lifetime for the fast-growing heterotrophic biofilms and a shorter fluorescence lifetime for the slow-growing autotrophic biofilms. Furthermore, starved biofilms showed shorter lifetimes than biofilms supplied with glucose, indicating a lower RNA:DNA ratio in starved biofilms. It is suggested that FLIM in combination with SYTO 13 represents a useful tool for the in situ differentiation of active and inactive bacteria. The technique does not require radioactive chemicals and may be applied to a broad range of sample types, including suspended and immobilized microorganisms.
In situ activity measurement represents an ongoing issue in various areas of applied and environmental microbiology. As early as 1985, Staley and Konopka reflected on different techniques employed to study microbial activity of single bacterial cells versus that of the bacterial community(36). In the meantime, several studies compared the results of different techniques for measuring microbial activity (3, 21, 33, 35, 36, 40). The major techniques comprise metabolism of radiolabeled substances, tetrazolium reduction (5, 9, 35, 39), and bromodeoxyuridine incorporation in DNA (1, 2, 12, 37). The most frequently used techniques are the ones using radioactive compounds either as a labeled nucleotide ([H3]thymidine) (34) or as a labeled carbon substrate (e.g., [H3]acetate). An elegant technique of activity measurement with simultaneous phylogenetic identification is fluorescent in situ hybridization combined with microautoradiography, which is becoming a widely used technique in different areas of microbiological research (6,11,14,20,28-32,43). Fluorescent in situ hybridization-microautoradiography samples are analyzed by confocal laser scanning microscopy (CLSM) running the instrument in both the fluorescent mode (“fluorochrome” emission of gene probe) and the reflection mode (“radioactive” signal of silver grains). CLSM is also well established as an in situ technique for studying microbial communities, especially bioaggregates and biofilms, without any fixation and embedding (18, 24). A relatively new technique for characterization of microbial communities is laser scanning microscopy with two-photon excitation (2P-LSM). 2P-LSM takes advantage of infrared excitation with a pico- or femtosecond pulsed laser source. So far, two-photon instruments have been used in only a few microbiological studies, e.g., studies of dense oral biofilms (42), marine stromatolites (15), assessment of fluorochromes for biofilm examination (22), differentiation of cyanobacteria and eukaryotic algae in phototrophic biofilms (27), and spatial distribution of zinc in microbial biofilms (13). A comparison of one-photon versus two-photon laser scanning techniques and their application in biofilm research has been discussed recently (25). Both CLSM and 2P-LSM are techniques recording the emission intensities of fluorochromes. However, the intensity is affected by several parameters, such as laser intensity, fluorochrome concentration, absorption, quenching, scattering, etc. Nevertheless, excited fluorochromes can deliver two pieces of information, first an intensity signal and second a lifetime signal. Fluorescence lifetime imaging microscopy (FLIM) is an emerging technique, and its potential applications have been discussed in several review articles (4, 7, 8). In a first microbiological application, FLIM was employed to measure pH gradients in densely grown artificial biofilms (42). The applicability of two-photon excitation using two-photon intensity and two-photon lifetime imaging in microbial ecology has been reported as well (26).
One commercially available FLIM technique measures the fluorescence lifetime by means of time-correlated single photon counting. The technique is based on a pulsed laser light source, e.g., a two-photon infrared laser, the detection time of photons, and the buildup of photon distribution versus time across the whole measurement period. This information is recorded at every single pixel of the image. From the photon distribution over time, the fluorescence decay curve can be recorded. The decay curve is then fitted to an exponential decay in order to determine the lifetime. Using this information, a color-coded fluorescence lifetime image can be calculated (10).
In this study, we employed FLIM as a tool for measuring microbial activity. The different levels of activity were determined by employing a nucleic acid-specific fluorochrome and its lifetime signal in dependency on binding to different types of nucleic acids. The basic idea is to take advantage of the fact that in actively growing bacteria, there is a higher concentration of RNA than in inactive bacteria. Consequently, the ratio of the two lifetimes, RNA:DNA, can be used as a measure in order to identify active and inactive bacterial cells. For this purpose, experimental systems at various levels of complexity were employed. Initial measurements were done with purified nucleic acids in solution, and then pure liquid bacterial cultures were analyzed at the single-cell level; finally, multispecies heterotrophic and chemoautotrophic biofilms were examined.
tRNA from Escherichia coli strain W (Sigma-Aldrich, St. Louis, MO) was solved in TE buffer (100 mM TRIS, 10 mM EDTA, pH 8) at a final concentration of 3,770 μg/ml. DNA from E. coli strain B (Sigma-Aldrich) was solved in TE buffer (100 mM TRIS, 10 mM EDTA, pH 8) at a final concentration of 31 μg/ml. The size of E. coli DNA was studied by using agarose gel electrophoresis and turned out to be >3,000 bp. In addition, DNA from the slow-growing BNC1 strain (17) was isolated using the DNeasy tissue kit (QIAGEN, Hilden, Germany). DNA purified by this procedure was typically up to 50 kb in size, with fragments of 30 kb predominating (DNeasy Tissue Handbook).
Escherichia coli strain K-12 (JM 83, no. 3947; German Collection of Microorganisms and Cell Cultures [DSMZ], Braunschweig, Germany) was grown in liquid LB medium at 37°C. For subculturing, 1 ml of an overnight culture was inoculated in 200 ml of medium in 500-ml Erlenmeyer flasks and shaken at 130 rpm until the culture was in the stationary phase. The cell number and concentration were established microscopically using a Thoma chamber (Hecht, Sondheim, Germany). During the growth curve of the bacteria, 1 ml of the culture was sampled every 30 or 45 min for subsequent FLIM measurements.
For each time point in the growth curves of bacterial monocultures, the growth rate (μ) was computed from the increase/decrease of the cell density to the next-nearest-neighbor points. The growth rate at time point t2 (μ) was obtained as follows. First, from the cell density at time t2 (cd2) and the cell density at the previous time point, t1 (cd1), the growth rate (μ1/2) was calculated according to the following equation: μ1/2 = [ln(cd2) − ln(cd1)]/(t2 − t1). Then, from the cell density at time t2 (cd2) and the cell density of the following time point t3 (cd3), the growth rate (μ2/3) was calculated. The final growth rate at time point t2 was obtained as the average of the two growth rates of μ1/2 and μ2/3.
Biofilms were grown in rotating annular reactors on polycarbonate or glass slides (19, 23). Sewage sludge taken from a municipal wastewater treatment plant (Hildesheim, Germany) was used as an inoculum. Chemoautotrophic biofilms were cultivated in SI-N1 trace element solution (0.5 g Na2EDTA, 3 g FeSO4·7H2O, 0.3 g Na2MoO4·2H2O, 0.2 g CuSO4·5H2O in 1,000 ml) with 1 g liter−1 MgSO4 and 0.4 g liter−1 CaCl2. A substrate load based on a biofilm surface area of 1.5 or 3 g m−2 day−1 ammonium (NH4+) and 1.5 g m−2 day−1 hydrogen carbonate (HCO3−) was used. For cultivation of heterotrophic biofilms, SL 6 trace element solution (0.3 g H3BO3, 0.13 g CoCl2, 0.014 g CuCl2·2H2O, 0.015 g MnSO4, 0.031 g Na2MoO4·2H2O, 0.01 g NiCl2·6H2O, 0.056 g ZnSO4·7H2O in 1,000 ml) with 1 g liter−1 MgSO4, 0.4 g liter−1 CaCl2, 1 g liter−1 NaNO3, and a substrate load of 1.5 g m−2 day−1 glucose was employed. The biofilm samples for FLIM were taken at days 11, 30, and 68 for both heterotrophic and autotrophic biofilms. Autotrophic biofilms were additionally sampled at day 106.
The nucleic acid-specific fluorochrome SYTO 13 (Molecular Probes, Eugene, OR) was used in all experiments. The samples were stained with SYTO 13 for 5 min in the dark at room temperature and at a final concentration of 10 μM in TE buffer.
Laser scanning microscopy was done with a TCS SP1 in combination with an upright microscope (Leica, Mannheim, Germany). The system was controlled by the Leica Confocal Software, version 2.5, build 1104. Images of samples were recorded using 63× 1.2-numerical-aperture and 63× 0.9-numerical-aperture water immersion/immersible lenses, respectively. For two-photon excitation, the setup was equipped with a triple-laser system consisting of two laser diode bars, a Millennia Vs continuous-wave pump laser, and a tunable mode-locked Ti/Sapphire Tsunami laser with the midwave mirror set (Spectra Physics, Mountain View, CA). A wavelength of 800 nm was used for two-photon excitation of SYTO 13. The fluorescence emission of SYTO 13 was detected in the range of 400 to 800 nm. The SPC-730 module working on a time-correlated single-photon counting principle and the SPC-730 software, version 8.5, were available for detection of time-resolved fluorescence (Becker & Hickl, Berlin, Germany). FLIM signals were detected using a PMH-100 detector for photon counting (Hamamatsu, Hamamatsu City, Japan) mounted to the RLD port (nondescanned detector) of the laser scanning microscope. Standard recording time for lifetime imaging was 60 s.
In a first step, the computation of fluorescence lifetimes and digital image analysis of FLIM images was performed using the SPC Image software, version 2.84 (Becker & Hickl). Then, the images were exported into the ImageJ software program (NIH; [http://rsb.info.nih.gov/ij/]). ImageJ was used with a self-written plugin for further analysis of the FLIM images. No background correction was performed, since the background fluorescence from the cultured strains and the autotrophic biofilms was found to be negligible (data not shown). In the bacterial pure culture experiments, each bacterium was manually selected with the region-of-interest (ROI) tool and the average fluorescence lifetime in nonzero pixels of this region was taken. For measurements in nucleic acid solutions, the average fluorescence lifetime from the whole image was evaluated in the case of tRNA and DNA from the BNC1 strain. In the case of DNA from E. coli, the precipitated DNA fibrils were chosen as the ROI and evaluated. In the biofilm experiments, ROIs were drawn around the characteristic spherical colonies of the autotrophic biofilms. In parallel, the average fluorescence lifetime from the whole image was evaluated.
The correlation between groups was proven by using the Spearman rank order correlation test. Differences between the groups was proven by using the unpaired t test (to compare two groups) or one-way analysis of variance followed by Dunn's method (to compare >2 groups). If the data were not normally distributed, the Mann-Whitney rank sum test (to compare 2 groups) or the Kruskal-Wallis analysis of variance on ranks (to compare >2 groups) was applied. The significance level in all tests was a P value of <0.05. Results are given as means ± standard deviations unless stated otherwise.
Fluorescence lifetimes of SYTO 13 in E. coli DNA and tRNA solutions and their mixtures were measured. This should simulate the situation in the living cell, where both nucleic acids are present. Thus, after staining with SYTO 13, the average fluorescence lifetime of SYTO 13-DNA complexes and SYTO 13-RNA complexes was obtained. The average fluorescence lifetime from 256-by-256 pixel images were computed for tRNA as well and the tRNA/DNA mixture. In the case of pure DNA, the average fluorescence lifetime in the precipitates seen in the images was assessed, since the bulk phase did not show sufficient fluorescence intensity for evaluation of lifetimes. The results (Fig. (Fig.1)1) showed shorter fluorescence lifetimes for the pure DNA solution (2,715 ± 121 ps; n = 22 images) than for the pure tRNA solution (3,590 ± 42 ps; n = 10 images). The fluorescence lifetime of SYTO 13 in the tRNA:DNA mixture (concentration ratio [weight:weight] of tRNA:DNA = 3.39:1) is situated in between the values for pure DNA and tRNA (3,212 ± 63 ps; n = 10 images). The average in vivo fluorescence lifetimes of SYTO 13-stained E. coli cells (2,739 ± 247 ps; n = 1,553 cells) lies close to that of the pure DNA solution.
In a first experiment with autotrophic biofilms, a very short lifetime of SYTO 13 was measured. The fluorescence lifetime of SYTO 13 in slow-growing, autotrophic biofilms was 1,853 ± 203 ps (n = 82 images) and thus was short compared to results of the in-solution experiments with E. coli-derived DNA and tRNA (see Fig. Fig.1).1). Therefore, we have performed additional experiments with SYTO 13- stained DNA solutions purified from the slow-growing bacterial strain BNC1 in order to check the possibility that the fluorescence lifetimes of SYTO 13 bound to DNA isolated from slow-growing bacterial strains is significantly shorter than those for E. coli-derived DNA. The measurements of SYTO 13-stained DNA purified from the slow-growing EDTA-degrading BNC1 strain showed an even shorter fluorescence lifetime (1,523 ± 293 ps; n = 13 images) if compared with DNA from E. coli.
The experiments with bacterial monocultures were the next step on the way to establishing the fluorescence lifetime measurements of bacterial growth and activity in complex microbial communities. During the growth curve of E. coli in liquid medium, cell counts were determined. The fluorescence lifetime of SYTO 13 was monitored every 45 min. The bacteria were inoculated in fresh medium and observed until the population reached the stationary growth phase.
SYTO 13 fluorescence lifetimes were assessed from fluorescence lifetime images using the freehand ROI tool. The average lifetime of each bacterium in the image was computed for images with <10 bacteria, and the average fluorescence lifetimes of 10 randomly selected bacteria were assessed for images with more than 10 bacteria. In E. coli liquid culture (Fig. (Fig.2),2), the lifetime was longer in the exponential growth phase than in the lag and stationary growth phases. Standard errors of means, which show the diversity of the lifetimes in the bacterial population, were smaller in the late exponential phase than in the lag and stationary phases. This correlates with the lower diversity of growth rates for the exponentially growing population than for populations in lag or stationary growth phases. The correlation between growth rates μ (Fig. (Fig.3)3) and the fluorescence lifetimes was significant in all experiments (4 experiments, 9 to 12 time points per experiment, and typically 20 to 50 bacteria per time point).
The fluorescence lifetimes of SYTO 13-stained chemoautotrophic and heterotrophic biofilms growing in rotating annular reactors were compared in order to study whether the lifetimes correlate with the growth rate of the biofilms (autotrophic biofilms have a lower growth rate than heterotrophic ones). Fluorescence lifetimes of SYTO 13 in autotrophic (slower-growing) biofilms were significantly shorter than those with heterotrophic biofilms (1,853 ± 203 ps [n = 82 images] for autotrophic biofilms; 2,486 ± 390 ps [n = 54 images] for heterotrophic biofilms). Furthermore, in the case of autotrophic biofilms, the effects of different substrate loads on the fluorescence lifetime were studied. Under a higher substrate load, which is supposed to result in higher growth rates of biofilm organisms, the biofilm showed longer fluorescence lifetimes, indicating higher RNA:DNA ratios in the cells than was the case with the biofilm under lower substrate loads. In order to be sure that we compared the similar bacterial species in these autotrophic biofilms, we evaluated separately the lifetimes of typical spherical microcolonies of bacteria growing in our autotrophic biofilms and compared them (see Fig. Fig.4).4). The fluorescence lifetimes of autotrophic colonies grown with 3 g m−2 day−1 ammonium were significantly longer than those of colonies grown with 1.5 g m−2 day−1 ammonium (2,377 ± 257 ps [n = 38 colonies] for higher substrate load; 1,977 ± 253 ps [n = 69 colonies] for lower substrate load).
Slides from the reactor experiments with heterotrophic and autotrophic biofilm media were removed from the reactor and incubated in petri dishes in buffer without nutrients for 24 h in order to evaluate the impact of starvation on the lifetime of SYTO 13. It is known that starvation decreases the RNA:DNA ratio. The control biofilms were left in the reactor (and supplied with glucose or ammonium) till 1 to 2 h before the measurements were taken. The starved heterotrophic biofilms showed significantly shorter fluorescence lifetimes than the control (2,562 ± 291 ps [n = 13 images] for starved biofilms; 2,970 ± 161 ps [n = 14 images] for glucose-supplied biofilms). For autotrophic biofilms, the difference in lifetimes between the starved and ammonium-supplied biofilms was not significant. Nevertheless, the mean values showed the same trend as in the case of heterotrophic biofilms, being shorter in case of starved autotrophic biofilms than with the ammonium-supplied autotrophic biofilms (1,814 ± 117 ps [n = 15 images] for starved biofilms; 1,848 ± 105 ps [n = 17 images] for ammonium-supplied biofilms).
The idea of using FLIM in combination with SYTO 13 for in situ bacterial activity measurement originates with a report on DNA and RNA discrimination in healthy and apoptotic Chinese hamster ovary cells (41). This approach was transferred to microbiological samples in order to employ it for measuring bacterial activity. The rationale of the experiments was that resting bacterial cells, e.g., those in the stationary growth phase, have a high DNA content and a low RNA content. During bacterial activity, e.g., in the logarithmic growth phase, the bacterial cells produce RNA and as a result the proportion of RNA/DNA changes. This change was determined by means of FLIM as a sum parameter or on the basis of single cells.
A first attempt at using the FLIM approach on biofilm samples showed that it is possible to distinguish bacteria exposed to different nutrient regimes (26). For this purpose, biofilms were developed with river water in a rotating annular reactor on polycarbonate slides. The slides from the reactor were then incubated in petri dishes with either old river water, fresh river water, or fresh river water with glucose added. Measurement of the SYTO 13 lifetime showed significantly different signals. As a result, the lifetimes of SYTO 13 were approximately 1,700 ps (starved bacteria), 2,000 ps (slow-growing bacteria), and 2,200 ps (fast-growing bacteria). These preliminary results (26) encouraged the present study with defined nucleic acids in solution, pure cultures of E. coli, and chemoautotrophic and heterotrophic biofilms.
The nucleic acid-specific fluorochrome SYTO 13 becomes fluorescent after binding RNA and DNA. For both single- and double-stranded nucleic acids, the quantum yields are equal. SYTO 13 has an excitation maximum at 488 nm and an emission peak at 510 nm. In comparison to the fluorescence emission intensity, the lifetime of SYTO 13 is significantly different if it is bound to RNA or DNA (Fig. (Fig.1).1). This result is in agreement with results of a previous study investigating the distribution of DNA and RNA in eukaryotic cell cultures. Van Zandvoort et al. could demonstrate different lifetimes for SYTO 13 bound to DNA (nucleus and mitochondria) and to RNA (cytoplasm) using differential imaging (41). The differences in the lifetime of isolated E. coli DNA in comparison to DNA in E. coli cells may be due to the local microenvironment. For example, the cellular DNA is associated with proteins which may slightly shift the lifetime measured. The lifetime also followed the growth curve measured for suspended E. coli cells (Fig. (Fig.2).2). If the growth rate of E. coli is calculated and compared with the lifetime measured, a significant correlation can be seen (Fig. (Fig.3).3). This strongly suggests the applicability of lifetime imaging as an in situ technique for measuring bacterial cell activity.
The RNA:DNA ratio was already discussed as a potential technique as a measure for estimating the growth rate of bacteria. However, the technique requires the extraction of nucleic acids in order to measure RNA and DNA using a spectrophotometer after staining with ethidium bromide (16). The discrimination of different bacterial populations in seawater was described using flow cytometry (38). Troussellier and colleagues were able to separate different bacterial clusters using the fluorescence intensity of SYTO 13. The emission intensity was taken as a measure of nucleic acid content and cell size. These clusters might represent either the same or different bacterial species. We have shown that in environmental biofilms, it is possible to differentiate bacterial populations based on their fluorescence lifetimes after staining with SYTO 13 (Fig. (Fig.4).4). When FLIM is used, the information about cell size and fluorescence intensity is intrinsically present or can be easily obtained from the fluorescence intensity image, which is always computed automatically prior to the fluorescence lifetime image. Consequently, the fluorescence intensity in combination with the fluorescence lifetime will reveal more information on environmental microbial populations. This combined approach allows the recording of both structural information on cellular and polymeric biofilm constituents and physiological information, e.g., bacterial activity.
The FLIM approach in combination with the nucleic acid-specific fluorochrome SYTO 13, described for eukaryotic cells elsewhere, can be transferred to microbiological samples.
The fluorescence lifetimes of SYTO 13 in tRNA and DNA solutions were in agreement with previous measurements, which confirmed the assumptions for measurement of RNA/DNA ratios.
The division rate [μ] of E. coli strongly correlates with the fluorescence lifetime of SYTO 13, showing the suitability of the method for measurement of cell activity in liquid bacterial monocultures.
The FLIM approach for RNA:DNA ratio measurements can easily be extended to environmental microbial communities, as suggested by experiments with slow- and fast-growing biofilm communities.
FLIM as a tool for measuring bacterial activity does not require extraction of RNA and DNA, no radioactive chemicals are necessary, and it can be applied as an in situ technique.
P.W. acknowledges financial support from UFZ.
We thank B. Nörtemann for the EDTA-degrading strain BNC1. The supply of reactor biofilms from Christian Staudt is appreciated. The plugin for ImageJ was written by M. Tröger and R. Braungarten.
Published ahead of print on 2 November 2007.